AI/ML●●●Banger
Opus Magnum Bench -- Shape Rotation and Alchemical Engineering
Game-based AI benchmark measuring spatial reasoning against human speedrun records.
Big BrainNiche Gem
ClassicRob
3027d ago
Back to browse
All the LM solutions on SWE-bench are bloated compared to humans
by lieret·Mar 4, 2026·1 point·0 comments
AI Analysis
○Pass
Twitter thread with a chart; not a product or tool.
Strengths
- •Clear visual comparison of patch bloat across multiple models.
- •Identifies real pattern in LLM code generation (verbose output).
Weaknesses
- •No actionable tool, framework, or code provided—pure analysis posted to Twitter.
- •No link to reproducible methodology or dataset beyond the thread.
Category
Target Audience
AI researchers, ML practitioners studying code generation models
Similar Projects
AI/ML●Mid
I pre-registered the rules,then measured AI agents shipping wrong fixes
Pre-registered methodology shows 17/50 wrong fixes ungated vs 0/50 gated, but the actual tool is private.
Big Brain
kolesnikov-arch
2114d ago
Developer Tools●●Solid
Codex context bloat? 87% avg reduction on SWE-bench Verified traces
Transparent proxy cuts Codex context tokens by 87% via working memory.
Big BrainNiche Gem
george_ciobanu
1022mo ago
AI/ML●●Solid
LLMxRay an open-source observability tool for LLMs
Multilingual tokenization comparison across Arabic, Chinese, French that LangSmith ignores.
Big BrainNiche Gem
lognebudo
104mo ago
AI/ML●●●Banger
97% on SWE-bench Verified with subscription-token agents
97% on SWE-bench Verified with full artifact transparency, not just a score claim.
Big BrainZero to One
kimjune01
201mo ago
AI/ML●●Solid
17MB model beats human experts at pronunciation scoring
Beats humans at pronunciation scoring but doesn't ship product integration yet.
Big BrainWizardry
fabiosuizu
1315mo ago